Exports and Productivity: A survey of the evidence from firm level data. Joachim Wagner. University of Lüneburg Working Paper Series in Economics

Transcription

1 Exports and Productivity: A survey of the evidence from firm level data Joachim Wagner University of Lüneburg Working Paper Series in Economics No. 4 Lüneburg, March ISSN

2 Exports and Productivity: A survey of the evidence from firm level data Joachim Wagner University of Lueneburg, Institute of Economics Institute for the Study of Labor (IZA), Bonn Hamburg Institute of International Economics (HWWA), Hamburg wagner@uni-lueneburg.de Prepared for a special issue of The World Economy on Exports and Growth [This version: March 2, 2005] Abstract: While the role of exports in promoting growth in general, and productivity in particular, has been investigated empirically using aggregate data for countries and industries for a long time, only recently have comprehensive longitudinal data at the firm level been used to look at the extent and causes of productivity differentials between exporters and their counterparts which sell on the domestic market only. This papers surveys the empirical strategies applied, and the results produced, in 45 microeconometric studies with data from 33 countries that were published between 1995 and Details aside, exporters are found to be more productive than non-exporters, and the more productive firms self-select into export markets, while exporting does not necessarily improve productivity. Keywords: Exports, productivity, literature survey, micro data JEL classification: F14, D21 Address for correspondence: Prof. Dr. Joachim Wagner University of Lueneburg Institute of Economics Campus D Lueneburg Germany Phone: Fax:

3 1. Motivation Discussions of the role of exports in promoting growth in general, and productivity in particular, have been ongoing for many years. Until some ten years ago, empirical studies in this field used data at the country or industry level to test whether exports promote productivity growth or vice versa (see the surveys by Baldwin (2000) and Giles and Williams (2000a, 2000b)). In 1995 Bernard and Jensen published the first of series of papers that changed this research perspective (see Bernard and Jensen 1995, 1999, 2004a). They used large comprehensive longitudinal data from surveys performed regularly by official statistics in the U. S. to look at differences between exporters and non-exporters in various dimensions of firm performance, including productivity. These papers started a literature. During the ten years following the publication of Bernard and Jensen s Brookings paper researchers all over the world discovered the rich data sets collected by their statistical offices as a source to investigate the export activity of firms, and its causes and consequences. 1 The extent and cause of productivity differentials between exporters and their counterparts which sell on the domestic market only is one of the core topics in this literature. There are two alternative but not mutually exclusive hypotheses why exporters can be expected to be more productive than non-exporting firms (see Bernard and Jensen 1999; Bernard and Wagner 1997): The first hypothesis points to self-selection of the more productive firms into export markets. The reason for this is that there exist additional costs of selling goods in foreign countries. The range of extra costs include transportation costs, distribution or marketing costs, personnel with skill to manage foreign networks, or production costs in modifying current 2

4 domestic products for foreign consumption. These costs provide an entry barrier that less successful firms cannot overcome. Furthermore, the behaviour of firms might be forwardlooking in the sense that the desire to export tomorrow leads a firm to improve performance today to be competitive on the foreign market, too. Cross-section differences between exporters and non-exporters, therefore, may in part be explained by ex ante differences between firms: The more productive firms become exporters. The second hypothesis points to the role of learning-by-exporting. Knowledge flows from international buyers and competitors help to improve the post-entry performance of export starters. Furthermore, firms participating in international markets are exposed to more intense competition and must improve faster than firms who sell their products domestically only. Exporting makes firms more productive. These two hypotheses are by no means mutually exclusive. This paper reviews the findings of studies that use micro data at the level of firms (i.e. plants, establishments, local production units) to investigate the relationship between export activities and productivity empirically. The rest of the paper is organised as follows: Section 2 describes the various empirical approaches used in the post-1995 literature to identify the extent and direction of the relationship between exports and productivity using micro data at the firm level. Section 3 reviews 45 studies with data from 33 countries and summarises the core results. Section 4 concludes. 1 Earlier research using longitudinal micro data from official statistics in Germany to investigate causes and consequences of exporting is summarised in Wagner (1995). 3

5 2. Investigating the exports / productivity relationship: Empirical strategies 2.1 A standard approach A common approach to investigate differences in productivity between exporters and nonexporters is to follow (sometimes only in part, and sometimes with modifications and extensions) the methodology introduced by Bernard and Jensen (1995, 1999). Studies of this type use longitudinal data for plants (usually from the regular surveys conducted by official statistics) to document differences in levels and growth rates of productivity between exporters and non-exporters in a first step. Here one starts by looking at differences in average labour productivity (total value of shipments per worker, or value added per worker) or average total factor productivity 2 between exporters and non-exporters. The result is an unconditional productivity differential. The next step is the computation of so-called exporter premia, defined as the ceteris paribus percentage difference of labour productivity between exporters and non-exporters. These premia are computed from a regression of log labour productivity on the current export status dummy and a set of control variables (usually including industry, region, firm size measured by the number of employees, and year): (1) ln LP it = a + ß Export it + c Control it + e it where i is the index of the firm, t is the index of the year, LP is labour productivity, Export is a dummy variable for current export status (1 if the firm exports in year t, 0 else), Control is a vector of control variables (like four-digit industry dummies, dummies for regions, firm size, and year dummies), and e is an error term. The export premium, computed from the estimated 2 To simplify the exposition we will refer to labour productivity only from now on. 4

6 coefficient ß as 100(exp(ß)-1), shows the average percentage difference between exporters and non-exporters controlling for the characteristics included in the vector Control. To control for unobserved plant heterogeneity due to time-invariant firm characteristics which might be correlated with the variables included in the empirical model and which might lead to a biased estimate of the exporter premia, a variant of (1) is often estimated with fixed plant effects, too. Next, differences in productivity growth between exporters and non-exporters are investigated based on an empirical model that can be written as (2) ln LP it - ln LP i0 = a + ß 1 Start it + ß 2 Both it + ß 3 Stop it + c Control i0 + e it where Control is a vector of plant characteristics in year 0, and the dummies for export status are defined as follows: Start it = 1 if (Export i0 = 0) and (Export it = 1) Both it = 1 if (Export i0 = 1) and (Export it = 1) Stop it = 1 if (Export i0 = 1) and (Export it = 0) where non-exporting in both years is the reference category. The regression coefficients ß 1, ß 2 and ß 3 are estimates for the increase in growth rates of labour productivity for export starters, exporters in both years, and export stoppers relative to non-exporters in both years, controlling for firm characteristics included in the vector Control. Here we look at ß 2 to compare exporters and non-exporters. 5

7 To shed light on the empirical validity of the first hypothesis mentioned namely, that the more productive firms go abroad the pre-entry differences in productivity between export starters and non-exporters are investigated next. If good firms become exporters then we should expect to find significant differences in performance measures between future export starters and future non-starters several years before some of them begin to export. To test whether today s export starters were more productive than today s non-exporters several years back when all of them did not export, select all firms that did not export between year t-3 and t-1, and compute the average difference in labour productivity in year t-3 between those firms who did export in year t and those who did not. More formally, estimate the empirical model (3) ln LP it-3 = a + ß Export it + c Control it-3 + e it where i is the index of the firm, t is the index of the year, LP is labour productivity in year t-3, Export is a dummy variable for current export status (1 if the firm exports in year t, 0 else), Control is a vector of control variables (like four-digit industry dummies, dummies for regions, firm size, and year dummies), and e is an error term. The pre-entry premium, computed from the estimated coefficient ß as 100(exp(ß)-1), shows the average percentage difference between today s exporters and today s non-exporters three years before starting to export, controlling for the characteristics included in the vector Control. To investigate the related question whether productivity increased more in export starters in the years before the start than in firms that continue not to export, the empirical model (4) ln LP it-1 - ln LP it-3 = a + ß Export it + c Control i0 + e it is used. The estimated regression coefficient ß shows the extent in which future exporters outperformed the non-exporting firms in the years prior to entry. 6

8 To test for the second hypothesis mentioned namely, that exporting fosters productivity - the post-entry differences in productivity growth between export starters and non-exporters are investigated. This is done by looking at b 1 from (2) to compare the productivity growth performance of export starters and non-exporters. Finally, to find out whether stopping to export is negatively related with productivity performance, post-exit differences in productivity growth between export stoppers and nonexporters are investigated by looking at b 3 from (2) to compare the productivity growth performance of export stoppers and non-exporters. While most of the empirical studies that use (variants of) the now standard approach outlined in this section compare exporters and non-exporters across all (manufacturing) industries, some focus on firms from selected industries only and document interesting similarities and differences (see e.g. Alvarez and López (2004), Blalock and Gertler (2004), De Loecker (2004), and Greenaway and Kneller (2004b)). Furthermore, Damijan, Polanec and Prasnikar (2004) recently looked at differences by foreign markets served and found that it matters whether firms exported to advanced countries or developing countries. 2.2 Extensions The standard approach outlined in the last section has been augmented by extensions and alternative approaches that deal with some of its weaknesses and problems. Here we will discuss two of these recent developments that are used more and more in empirical investigations, namely the comparison of productivity between matched firms, and differences in the distribution of productivity as a whole between exporters and non-exporters. 7

9 To motivate the first approach mentioned, consider the following situation: Assume that a study reports that plants entering the export market have substantially faster productivity growth in the following years than firms that keep selling their products on the domestic market only. Does this point to a causal effect of starting to export on productivity? The answer is, obviously, no: If better firms self-select into export-starting, and if, therefore, todays export starters are 'better' than today s non-exporters (and have been so in the recent past), we would expect that they should, on average, perform better in the future even if they do not start to export today. However, we cannot observe whether they would really do so because they do start to export today; we simply have no data for the counterfactual situation. So how can we be sure that the better performance of starters compared to non-exporters is caused by exporting (or not)? This closely resembles a situation familiar from the evaluation of active labor market programs (or any other form of treatment of units): If participants, or treated units, are not selected randomly from a population but are selected or self-select according to certain criteria, the effect of a treatment cannot be evaluated by comparing the average performance of the treated and the non-treated. However, given that each unit (plant, or person, etc.) either participated or not, we have no information about its performance in the counterfactual situation. A way out is to construct a control group in such a way that every treated unit is matched to an untreated unit that has been as similar as possible (ideally, identical) at the time before the treatment. Differences between the two groups (the treated, and the matched non-treated) after the treatment can then be attributed to the treatment (for a comprehensive discussion, see Heckman, LaLonde and Smith 1999). The use of a matching approach to search for causal effects of starting or stopping to export on productivity (and other dimensions of firm performance) has been pioneered by Wagner (2002) and Girma, Greenaway and Kneller (2003, 2004), and it has been used in a growing 8

10 number of empirical studies ever since (including De Loecker (2004), Arnold and Hussinger (2004), and Alvarez and López (2004)). As regards the second recent methodological innovation in this literature, consider the comparison of productivity (or productivity growth) between exporters and non-exporters. If one looks at differences in the mean value for both groups only, one focuses on just one moment of the productivity distribution. A stricter test that considers all moments is a test for stochastic dominance of the productivity distribution for exporters over the productivity distribution for non-exporters. More formally, let F and G denote the cumulative distribution functions of productivity for exporters and non-exporters. Then first order stochastic dominance of F relative to G means that F(z) G(z) must be less or equal zero for all values of z, with strict inequality for some z. Whether this holds or not is tested non-parametrically by adopting the Kolmogorov-Smirnov test. This method has been used to discuss the issue of exports and productivity for the first time by Delgado, Farinas and Ruano (2002); recent applications comparing firms that produce for the local market only, that export, and that are foreign direct investors are Girma, Kneller and Pisu (2003) and Girma, Görg and Strobl (2004). A related extension of the standard approach used in the investigation of the relationship between exports and productivity is the application of quantile regression, introduced to this field of analysis by Yasar, Nelson and Rejesus (2003). By construction this method examines the productivity effect of exporting at different points of the conditional output distribution. To state it differently, quantile regression allows to test for differences in the effects of exporting on plant productivity as one moves from the lower to the upper tail of the conditional productivity distribution, and to identify the regions where these effects are especially weak, or strong, or not significantly different from zero at all. 9

11 3. A survey of the evidence, During the ten years following the publication of the path-breaking Brookings paper by Bernard and Jensen (1995) researchers all over the world used firm level data to investigate the relationship between exporting and productivity in microeconometric studies. Table 1 gives a synopsis of findings from 45 empirical studies covering 33 countries. Among the countries covered are highly industrialised countries (e.g., U.S., UK, Canada, Germany); countries from Latin America (Chile, Colombia, Mexico); Asian countries (China, Korea, Indonesia, Taiwan); transition countries (Estonia, Slovenia); and least developed countries from sub-saharan Africa. Given this wide range of countries the big picture emerging from column two of table 1 that summarises findings on differences in levels and growth rates between exporters and nonexporters is amazingly clear-cut: With only a few exceptions exporters are found to have higher productivity, and often higher productivity growth, and this tends to hold after controlling for observed plant characteristics (like industry and size), too. Exporters are better. The findings for pre-entry differences surveyed in column three often present evidence in favour of the self-selection hypothesis: Future export starters tend to be more productive than future non-exporters years before they enter the export market, and often have higher ex-ante growth rates of productivity. The good firms go abroad. Evidence regarding the learning-by-exporting hypothesis is somewhat more mixed: Results for post-entry differences in performance between export starters and non-exporters collected in column four point to faster productivity growth for the former group in some studies only. 10

12 If matched firms are compared, often no statistically significant exporter premia are found. Exporting does not necessarily improve firms. Finally, a look at the results for post-exit differences collected in the last column reveals that stopping to export tends to be accompanied by a decrease in productivity in the most cases. Girma, Greenaway and Kneller (2003) in the only study using matched firms, however, find only weak negative effects in the year of exit, and no effect for later years. Obviously the big picture sketched here exporters are more productive than non-exporters, and the more productive firms self-select into export markets, while exporting does not necessarily improve productivity - hides a lot of cross-country heterogeneity which is documented in some detail in table 1, and in even more detail in the studies surveyed. Crosscountry comparisons, and even cross-study comparisons for one country, are difficult because the studies differ in details of the approach used. Therefore, the jury is still out on many of the issues regarding the relationship between exporting and productivity. One promising approach to generate stylised facts in a more convincing way is to co-ordinate microeconometric studies for many countries ex-ante, and to agree on a common approach and on the specification of the empirical models estimated. The outcome of such a joint effort would be a set of results that could be compared not only qualitatively (i.e. with regard to the signs and the statistical significance of the estimated coefficients) but with a view on the magnitude of the estimated effects, too. 3 3 See Bernard, Jensen and Wagner (1997) for this type of study using data for the U.S. and Germany. Volunteers willing to participate in an international study of this kind are asked to contact me! 11

13 4. Concluding remarks Details aside the big picture that emerges after ten years of microeconometric research in the relationship between exporting and productivity is that exporters are more productive than non-exporters, and that the more productive firms self-select into export markets, while exporting does not necessarily improve productivity. However, given all the difficulties (mentioned above) to compare the results from the vast numbers of studies in detail, it still seems to be too early to speak of these findings as stylised facts, and to discuss any policy conclusions to be based thereon. Furthermore, there are a number of important issues that have only been touched upon recently in some studies, and that deserve future research efforts that cover more countries: - If high-productivity firms self-select into export markets, is their high productivity due to an exogeneous random shock, or is it the results of a planned strategy to prepare for entering export markets? Hallward-Driemeier, Iarossi and Sokoloff (2002), López (2003), and Alvarez and López (2004) present evidence for the latter view based on data from five East Asian countries and Chile. - If exporting improves productivity via technology transfer from international buyers, what are the mechanisms by which this learning from exporting occurs? Blalock and Gertler (2004) report some anecdotal evidence from interviews with Indonesian exporting factory managers on this. - Which role is played by different target countries of exports for higher productivity as a precondition or result of exporting? Damijan, Polanec and Prasnikar (2004) report that in 12

14 Slovenia the productivity difference between future export starters and non-exporters is higher for firms that start to export to more advanced markets. To answer these questions, microeconometric research based on large sets of longitudinal plant level data has to be supplemented by field research in firms, following Susan Helper s (2000) credo that you can observe a lot just by watching. Case studies of this kind can not only produce the anecdotal evidence that helps us to understand what is behind the estimated coefficients that we produce with our PCs, they can point to the tailor-made questions to be included in future surveys that are aimed to collect data for a new generation of microeconometric studies, too. Furthermore, there is a different area of future research that is driven by an emerging theoretical literature. While at the dawn of the empirical literature surveyed here Leamer and Levinson (1994, p.1) stated that (i)nternational microeconomics is primarily a theoretical enterprise that seems little affected by empirical results, this is no longer true for some years now. A number of theoretical papers, including Bernard, Eaton, Jensen and Kortum (2003), Melitz (2003), Helpman, Melitz and Yeaple (2004), and Yeaple (2005), take the results from the empirical literature on firms and exports as a starting point and develop models of international trade with heterogeneous firms which focus on the relationship between productivity and exports. These theoretical models in turn generate testable hypotheses, and serve as catalysts for future microeconometric studies. 13

15 References Alvarez, Roberto and Ricardo A. López Exporting and Performance: Evidence from Chilean Plants. University of Chile and Indiana University, mimeo, February (forthcoming, Canadian Journal of Economics, 2005). Arnold, Jens Martin and Katrin Hussinger Export Behavior and Firm Productivity in German Manufacturing. A Firm-level Analysis. Centre for European Economic Research Discussion Paper Aw, Bee Yan, Xiaomin Chen and Mark J. Roberts Firm-level Evidence on Productivity Differentials, Turnover, and Exports in Taiwanese Manufacturing. Pennsylvania State University mimeo, June. Aw, Bee Yan, Sukkyun Chung and Mark J. Roberts Productivity and Turnover in the Export Market: Micro-level Evidence from the Republic of Korea and Taiwan (China). The World Bank Economic Review 14, Aw, Bee Yan and A. R. Hwang Productivity and the export market: A firm-level analysis. Journal of Development Economics 47, Baldwin, John R. and Wulong Gu Export-market participation and productivity performance in Canadian manufacturing. Canadian Journal of Economics 36, Baldwin, Robert Trade and Growth: Still Disagreement about the Relationships. Organisation for Economic Co-operation and Development, Economics Department Working Paper ECO/WKP(2000)37, October. Bernard, Andrew B Exporters and Trade Liberalization in Mexico: Production Structure and Performance. MIT mimeo, February. Bernard, Andrew B., Jonathan Eaton, J. Bradford Jensen and Samuel Kortum Plants and Productivity in International Trade. American Economic Review 93,

16 Bernard, Andrew B. and J. Bradford Jensen Exporters, Jobs, and Wages in U.S. Manufacturing: Brookings Papers on Economic Activity: Microeconomics Bernard, Andrew B. and J. Bradford Jensen Exceptional exporter performance: cause, effect, or both? Journal of International Economics 47, Bernard, Andrew B. and J. Bradford Jensen. 2004a. Exporting and Productivity in the USA. Oxford Review of Economic Policy 20, Bernard, Andrew B. and J. Bradford Jensen. 2004b. Why some firms export. Review of Economics and Statistics 86, Bernard, Andrew B., J. Bradford Jensen and Joachim Wagner The Good go Abroad: Evidence from Longitudinal Micro Data on German and U. S. Exporters. in Seppo Laaksonen (Ed.), The Evolution of Firms and Industries International Perspectives. Helsinki: Statistics Finland, Bernard, Andrew B. and Joachim Wagner Exports and Success in German Manufacturing. Weltwirtschaftliches Archiv / Review of World Economics 133, Bernard, Andrew B. and Joachim Wagner Export Entry and Exit by German Firms. Weltwirtschaftliches Archiv / Review of World Economics 137, Bigsten, Arne et al Exports and Firm-level Efficiency in African Manufacturing. Centre for the Study of African Economies, Oxford University, WPS/ , July. Blalock, Garrick and Paul J. Gertler Learning from exporting revisited in a less developed setting. Journal of Development Economics 75, Castellani, Davide Export Behavior and Productivity Growth: Evidence from Italian Manufacturing Firms. Weltwirtschaftliches Archiv / Review of World Economics 138,

17 Clerides, Sofronis K., Saul Lach and James R. Tybout Is Learning by Exporting Important? Micro-dynamic Evidence from Colombia, Mexico, and Morocco. Quarterly Journal of Economics CXIII, Damijan, Jose P., Saso Polanec and Janez Prasnikar Self-selection, Export Market Heterogeneity and Productivity Improvements: Firm Level Evidence from Slovenia. Katholieke Universiteit Leuven, LICOS Discussion Paper 148/2004, May. De Loecker, Jan Do Exports Generate Higher Productivity? Evidence from Slovenia. Katholieke Universiteit Leuven, LICOS Discussion Paper 151/2004, July. Delgado, Miguel A., Jose C. Farinas and Sonia Ruano Firm productivity and export markets: a non-parametric approach. Journal of International Economics 57, Farinas, José C. and Ana Martin-Marcos Exporting and Economic Performance: Firm-Level Evidence for Spanish Manufacturing. Universidad Complutense and UNED, Madrid, mimeo, April. Giles, Judith A. and Cara L. Williams. 2000a. Export-led growth: a survey of the empirical literature and some non-causality results. Part I. Journal of International Trade & Economic Development 9, Giles, Judith A. and Cara L. Williams. 2000b. Export-led growth: a survey of the empirical literature and some non-causality results. Part II. Journal of International Trade & Economic Development 9, Girma, Sourafel, Holger Görg and Eric Strobl Exports, international investment, and plant performance: evidence from a non-parametric test. Economics Letters 83, Girma, Sourafel, David Greenaway and Richard Kneller Export market exit and performance dynamics: a causality analysis of matched firms. Economics Letters 80,

18 Girma, Sourafel, David Greenaway and Richard Kneller Does Exporting Increase Productivity? A Microeconometric Analysis of Matched Firms. Review of International Economics 12, Girma, Sourafel, Richard Kneller and Mauro Pisu Exports versus FDI: An Empirical Test. University of Nottingham, GEP Research Paper 2003/21. Greenaway, David, Joakim Gullstrand and Richard Kneller Exporting May Not Always Boost Firm Level Productivity. University of Nottingham, GEP Research Paper 2003/26. Greenaway, David and Richard Kneller Exporting, Productivity and Agglomeration: A Difference in Difference Analysis of Matched Firms. University of Nottingham, GEP Research Paper 03/45. Greenaway, David and Richard Kneller. 2004a. Exporting and Productivity in the United Kingdom. Oxford Review of Economic Policy 20, Greenaway, David and Richard Kneller. 2004b. Industry Differences in the Effect of Export Market Entry: Learning by Exporting? University of Nottingham, GEP Research Paper 04/33. Greenaway, David and Zhihong Yu Firm-Level Interactions between Exporting and Productivity: Industry-Specific Evidence. Review of World Economics / Weltwirtschaftliches Archiv 140, Hahn, Chin Hee Exporting and Performance of Plants: Evidence from Korean Manufacturing. NBER Working Paper 10208, January. Hallward-Driemeier, Mary, Giuseppe Iarossi and Kenneth L. Sokoloff Exports and Manufacturing Productivity in East Asia: A Comparative Analysis with Firm-level Data. NBER Working Paper 8894, April. 17

19 Hansson, Pär and Nan Nan Lundin Exports as an Indicator on or Promoter of Successful Swedish Manufacturing Firms in the 1990s. Review of World Economics / Weltwirtschaftliches Archiv 140, Heckman, James J., R. J. LaLonde and J. A. Smith The economics and econometrics of active labor market programs. In: Ashenfelter, O. C. and David Card (Eds.), Handbook of Labor Economics, Vol. 3A, Amsterdam: North-Holland, Helper, Susan Economists and Field Research: You Can Observe a Lot Just by Watching. American Economic Review Papers and Proceedings 90, Helpman, Elhanan, Marc J. Melitz and Stephen R. Yeaple Export Versus FDI with Heterogeneous Firms. American Economic Review 94, Isgut, Alberto E What s Different about Exporters? Evidence from Colombian Manufacturing. Journal of Development Studies 37, Jensen, J. Bradford and Nathan Musick Trade, Technology, and Plant Performance. Economics and Statistics Administration, Office of Policy Development, ESA/OPD 96-4, April. Kraay, Aart Exports and economic performance: evidence from a panel of Chinese enterprises. In Mary-Francoise Renard (Ed.), China and its Regions. Economic Growth and Reform in Chinese Provinces. Cheltenham etc.: Elgar, (Originally published in Revue d Economie du Développement, no.1-2, 1999) Leamer, Edward E. and James Levinsohn International Trade Theory: The Evidence. NBER Working Paper Liu, Jin-Tan, Meng-Wen Tsou and James K. Hammitt Export Activity and Productivity: Evidence from the Taiwan Electronics Industry. Weltwirtschaftliches Archiv / Review of World Economcis 135, López, Ricardo A Entry to Export Markets and Firm-Level Productivity in Developing Countries. University of California, Los Angeles, mimeo, February. 18

20 Melitz, Marc J. The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity. Econometrica 71, Meller, Patricio Chilean Export Growth, : An Assessment. In G. K. Helleiner(Ed.), Manufacturing for Export in the Developing World. Problems and Possibilities. London and New York: Routledge, Mengistae, Taye and Catherine Pattillo Export Orientation and Productivity in Subsaharan Africa. IMF Staff Papers 51, Sinani, Evis Export market participation: The importance of sunk costs and spillovers. Center for East European Studies, Copenhagen Business School, mimeo, March. Sjöholm, Frederic Exports, Imports and Productivity: Results from Indonesian Establishment Data. World Development 27, Tsou, Meng-Wen, Jin-Tan Liu and James Hammitt Exporting and Productivity. Harvard School of Public Health mimeo, December. van Biesebroeck, Johannes Exporting Raises Productivity in Sub-Saharan African Manufacturing Plants. NBER Working Paper 10020, October. Wagner, Joachim Exports, Firm Size, and Firm Dynamics. Small Business Economics 7, Wagner, Joachim The causal effect of exports on firm size and labor productivity: First evidence from a matching approach. Economics Letters 77, Yasar, Mahmut, Carl H. Nelson and Roderick Rejesus Productivity and Exporting Status of Manufacturing Firms: Evidence from Quantile Regression. Emory University Department of Economics Working Paper 03-23, October. Yeaple, Stephen Ross A simple model of firm heterogeneity, international trade, and wages. Journal of International Economics 65,

21 Table 1: Synopsis of findings from empirical studies on exports and productivity using firm level data Core findings Study (published) Differences in levels and Pre-entry Post-entry Post-exit Country growth rates between differences differences differences (period covered) exporters and non-exporters Baldwin and Gu LP and TFP higher for exporters LP and growth of LP higher for Entrants had faster LP growth Exiters 13% less productive than (2003) than for non-exporters; difference starters than for non-exporters than non-entrants. continuers; exiters had slower growth Canada increased over time. EP for LP of LP than continuers. ( ) increased from 19 to 83 percentage points between 1974 and Meller LP significantly higher in small and (1995) large export firms than in non-export Chile firms; productivity differential ( ) differs between industries. Alvarez and López EP positive and significant for LP Firms that enter export have higher Differences in LP and TFP growth (2004) and TFP. Productivity differentials LP and TFP than non-exporters. insignificant or negative for export Chile differ considerably by industry. Firms make conscious efforts to starters compared to non-exporters. ( ) increase productivity before starting Same result for all industry but wood to export. Products. For matched firms no effect on TFP growth, weak positive for LP growth. Kraay LP and TFP significantly higher in For new entrants to export markets (2002) exporters than in non-exporters. EP learning effects are insignificant and China positive and significant for LP and occasionally negative. ( ) TFP. For established exporters, past exports are positively related to LP and TFP today, controlling for past firm performance and unobserved firm characteristics. Clerides, Lach and LP higher for exporting firms than LP higher for export starters LP improves after starting to LP shows worst performance compared Tybout (1998) for non-exporters. than for other groups of firms export. to other groups, particularly around time Colombia (ongoing exporters, non- of exit. ( ) exporters and export stoppers). 20

22 Isgut LP higher for exporting firms than Pre-entry premia 20% to 24%. LP growth rate differences Growth of LP not significantly lower for (2001) for non-exporters, 80% - 100% Pre-entry growth 3% - 4%. between export starters and non- stoppers compared to non-exporters Colombia for plants up to 100 employees and higher in future starters. exporters not significant for one over five year horizon. ( ) 27% - 32% for larger plants. EP ca. year horizon; LP grows 1.5% 45%. Growth of LP not significantly faster for starters over horizon different for exporters and non- five years after entry. exporters over five year horizon. Sinani LP higher for exporting firms than (2003) for non-exporters; growth of LP Estonia much higher for exporters than for ( ) non-exporters. Bernard / Wagner LP 3-4 % lower in smaller export Pre-entry premia 2-5 % but insign. Growth of LP significantly Growth of LP significantly lower (1997) firms, but percent higher in Pre-entry growth % higher for export starters for export stoppers than for non- Germany larger export firms. EP about 20 % higher in future starters, but than for non-exporters in the exporters in the year after exporting ( ) on average, increasing with share of difference insignificant. year after exporting starts stops ( %). exports in total sales. Growth of LP ( %). slower in exporting than in nonexporting firms. Bernard / Wagner Higher productivity strongly (2001) positively correlated with future Germany export entry, controlling for unob- ( ) served firm effects. Wagner EP in year before start positive but Growth of LP in export starters (2002) insignificant. higher than in matched non-starters Germany but difference between both groups ( ) not statistically significant. Arnold and Hussinger TFP higher for exporters than for In the two periods preceding entry Productivity gap between exporters (2004) non-exporters; high-productivity future exporters experience signific. and non-exporters does not widen Germany firms significantly more likely to increase in TFP. Productivity in years after entry. Exporting does not ( ) be exporters, ceteris paribus. granger-causes exporting. granger-cause productivity. For matched firms there are no differences in levels or growth of TFP between exporters and non-exporters in years after entry. 21

23 Sjöholm LP higher for exporting firms (1999) than for non-exporters; growth Indonesia of LP higher for exporters and ( ) increasing with share of exports in output. Blalock and Gertler LP higher for exporting firms Productivity does not rise prior to Firms experience a jump in No reduction in productivity after (2004) than for non-exporters. exporting. productivity of about 2% to 5% stopping to export. Indonesia following the initiation of exporting. ( ) Effect of exporting is positive in all 10 industries, significant in 7. Castellani LP higher for exporting firms LP higher in future export starters (2002) than for non-exporters; growth than in non-starters three years before Italy of LP not significantly different. entry; growth of LP not different for ( ) Productivity growth higher in the two groups of firms. LP and firms with a higher share of growth of LP have no impact on the exports in total sales. probability to start to export, cet. par. Girma, Görg and Strobl LP on average higher for (2004) exporters than non-exporters, Ireland but the hypothesis of identical (2000) distribution of productivity cannot be rejected for exporters relative to non-exporters. Aw, Chung and Total factor productivity (TFP) Higher TFP levels and growth Entrants have higher TFP than Exiting plants have higher TFP Roberts (2000) between 3.9% and 31.1% higher rates for entrants prior to entry non-exporters. than non-exporters in two industries; Korea (Republic of) for exporters than non-exporters not significant. no difference in three industries. ( ) in five industries. TFP growth not different between exporters and non-exporters. Hahn LP and TFP higher for exporters LP higher for entrants prior to Starters widen TFP gap with never Stoppers show decrease in TFP, (2004) than for non-exporters. EP about entry than for non-entrants, but no exporters and close gap with ever absolute and relative to ever Korea 50% - 20% for LP, 2.5% 7.5% for difference for TFP. No strong exporters. Effect pronounced in exporters, starter, and never TFP. evidence for difference in growth period after entry. exporters, before and after exit. rates of productivity ex ante. 22

24 Bernard LP almost 30% (shipments) Growth of LP not significantly Growth of LP not significantly (1995) or more than 50% (value added) different for export-starters and different for export stoppers and- Mexico greater for exporters. non-exporters. non-exporters. ( ) EP 34 % (value added). Growth of LP not significantly different for exporters and nonexporters. Clerides, Lach and LP higher for exporting firms than LP not higher for export starters No suggestion of a learning LP shows worst performance compared Tybout (1998) for non-exporters, export than for non-exporters and effect from exporting. to other groups. Mexico starters and export stoppers. lower than for exporters. ( ) Clerides, Lach and LP higher for exporting firms than LP higher for export starters LP improves after starting to LP sinks after stopping to export Tybout (1998) for non-exporters. than for non-exporters. export. Morocco ( ) Damijan, Polanec and Productivity of exporters higher Productivity in starters higher than No continuous productivity Firms ceasing exporting exhibit lower Prasnikar (2004) than of non-exporters; firms that in non-starters in years before improvement from exporting but productivity levels than old exporters Slovenia export to more markets are on starting. Productivity difference short run gains, only from serving up to 20%. ( ) average more labor productive. higher for firms that start to export advanced, high-wage foreign to more advanced markets. markets. De Loecker EP about 30% for value added For matched firms starting to export (2004) per worker. raises productivity instantly and also Slovenia in the years following. Analyses by ( ) industry find positive effects for most sectors, but these are significant in about half of them only. Delgado, Farinas TFP distribution for exporters TFP distribution for export No evidence of divergence of and Ruano (2002) stochastically dominates the starters stochastically dominates distribution of TFP growth Spain distribution for non-exporters the distribution for non-exporters between new exporters and non- ( ) prior to entry exporters; but post-entry growth greater for young entering exporters compared to young non-exporters. 23

25 Farinas and LP and TFP higher for exporters LP ex-ante higher for entering LP of entering exporters significantly LP of exiting exporters not different Martin-Marcos (2003) than for non-exporters. EP 17%. exporters than for continuing higher than LP of non-exporters. from non-exporters; dito for rate of Spain non-exporters. Growth of LP and TFP not different growth of LP and TFP. ( ) between entering exporters and continuing non-exporters. Greenaway, Gullstrand LP higher for exporters than for TFP lower for starters in the year For matched firms first time entry and Kneller (2003) non-exporters; TFP lower for of entry than for never-exporters. into export markets is not associated Sweden exporters, but around 10% higher with faster TFP growth compared to ( ) after controlling for industry fixed non-exporters. effects. Hansson and Lundin Productivity higher for exporters LP and TFP higher for future starters No significant differences in TFP (2004) than for non-exporters: EP 6.3% two years before entry, but lower growth between various export groups Sweden for TFP. (not significant) three years before. and non-exporters. Starters LP growth ( ) Differences in growth of TFP and LP higher than non-exporters. not significant. Aw / Hwang LP 36% greater for export-oriented (1995) than domestic-market-oriented firms Taiwan in electronics industry; differences (1986) vary between products examined. Aw, Chen and Higher total factor productivity Firms that eventually entered There may be some productivity Firms exiting the export market Roberts (1997) for exporting firms relative to non- the export market were more improvement associated with have higher productivity than Taiwan exporters from 11% in basic metals productive than their non-entering exporting. non-exporters. ( ) to 24% in textiles. counterparts in the years prior to their entry. Liu, Tsou and LP higher for exporters than non- Pre-entry growth ca. 8% - 12% Growth of LP substantially higher Growth of LP slower in export- Hammitt (1999) exporters in electronics industry. higher in future starters. for export starters than for non- stoppers than in non-exporters, Taiwan EP about 15%, and increasing with exporters (6.9% 8.7%). but coefficients not significant. ( ) export share. LP growth not different for exporters and nonexporters. 24

26 Aw, Chung and Total factor productivity (TFP) Entrants have higher TFP prior to Entrants are 13.3% to 18.9% more Exiting plants have average TFP Roberts (2000) between 11.8% and 27.6% higher entry than non-exporters. Initial productive than non-exporters. levels 4.4% to 10.3% higher than Taiwan for exporters than non-exporters difference widens after entry in non-exporters. Plants that exit fall ( ) in five industries. TFP growth in three industries. further behind exporting plants in three industries not different for the years following exit (significant exporters and non-exporters, in three industries). lower in two industries. Tsou, Liu and Growth of LP significantly higher Growth of TFP substantially higher Growth of TFP not different between Hammitt (2002) for plants that export across all for export starters than for non- stoppers and non-exporters. Taiwan three census periods compared to exporters. ( ) non-exporters, but result sensitive to cyclical patterns: little difference in downturn; exporters outperform nonexporters in upturn period. Yasar, Nelson and EP around 19% (OLS regression). Productivity about 23% higher for Productivity about 17% higher Rejesus (2003) EP vary significantly from 9% to entrants compared to non-exporters in stoppers compared to non- Turkey 21% from lower quantile to higher (OLS). Difference varies from 11% exporters (OLS). Difference ( ) quantile (quantile regression). to 21% between lowest and highest varies from 7% to 21% between quantile (quantile regression). lowest and highest quantile. Girma, Greenaway For matched firm exit has negative and Kneller (2003) albeit weak effect on TFP in the year UK ( ) of exit; no effect detected for later years. Girma, Greenaway Productivity higher for exporters Entrants more productive For matched firms: On entry year, and Kneller (2004) than for non-exporters. before entry than non-entrants. exporters experience TFP growth rate UK about 1.6 percentage points higher ( ) than non-starters. TFP continues to grow by an extra percentage point in the following year. Increase in share of exports raises rate at which TFP grows after entry. Greenaway and Productivity of exporters 5.4% For matched firms entry is Kneller (2003) above industry mean, of non- associated with significant increase UK exporters 4.6% below the mean. in LP. No robust evidence of ( ) productivity effects beyond the first few years for all firms, but for firms more exposed to export markets. 25

27 Greenaway and LP 2.2%, TFP 9.7% higher for Past TFP positive effect on entry. For unmatched firms, TFP growth Kneller (2004a) for exporters compared to non- Past productivity growth of future faster in years of and after entry than Exporters; EP 11.4% for LP and entrants higher compared to non- for continuing non-exporters. For 8.3% for TFP. entrants. matched firms differences are lower and only significant in entry year. Greenaway and Kneller Ceteris paribus the probability of For matched firms productivity (2004b) export entry is increasing in the growth in new export firms is on UK level of FTP. average 2.9% faster than in non- ( ) export firms. Effect is consistently lower in industries in which existing exposure to foreign firms is greater. Greenaway and Yu Exporters more productive than Higher TFP leads to higher Learning-by-exporting effect strongest (2004) non-exporters; EP 10.4% (output exporting probability. among new entrants, weaker for firms UK (Chemical Industry) per worker) and 9.1% (TFP). with more past export experience and ( ) negative for established exporters. Bernard / Jensen LP approx. a third greater for (1995) exporters across all plant sizes. U. S. EP about 15%. ( ) Jensen / Musick EP 13% in 1987 and 14 % in 1992 Growth of LP not significantly Growth of LP not significantly (1996) Growth of LP not significantly different for export-starters and different for export-stoppers and U. S. different for exporters and non- non-exporters. non-exporters. ( ) exporters. Bernard / Jensen EP ca. 20% (1984), ca. 16% Pre-entry premia 7% - 8%. Growth of LP significantly higher Growth of LP significantly lower (1999) (1987), ca. 18% (1992). Pre-entry growth in future starters for export starters than non- in export stoppers than in non- U. S. Short run: Higher growth rate of generally not statistically signif. exporters in the short, medium exporters in the short, medium ( ) LP in exporters than in non- different compared to non-starters. and long run. and long run. exporters; long run: no difference. Bernard / Jensen Plants that always export 8%-9% Two years before starting entrants In the year that they enter starters Plants that exit the export market have (2004a) more productive than plants that have productivity levels significantly have significantly faster productivity productivity growth rates 0.2%-0.9% U.S. never export. Exporters have 0.72% above continuing non-exporters, but growth rates than other firms. lower than continuing non-exporters. ( ) lower productivity growth rates per significantly below continuing year than similar plants producing exporters. solely for domestic market. 26

28 Bernard and Jensen More productive plants have higher (2004b) probability of starting to export, but U.S. controlling for plant fixed effects ( ) soaks the effect. Productivity effect is even negative (though insignificant) in GMM-first difference specification. Bigsten et al. (2000) Exporters exhibit higher average Initial exporters tend to exhibit Exporting in one period raises Cameroon ( ) efficiency levels than non- significantly higher levels of efficiency in the next period; the Ghana ( ) exporters. efficiency than other firms. first year of exporting raises Kenya ( ) efficiency by 14%. Zimbabwe ( ) Hallward-Driemeier, TFP larger for exporters than non- Iarossi and Sokoloff exporters; gap is larger the less (2002) developed is the local market. Firms Indonesia, Korea, that export from the beginning have Malaysia, Phillipines, higher levels of TFP years later, due Thailand ( ) to different firm policy (investment in fixed and human capital etc.). Van Biesebroeck EP for LP about 50%. LP higher for export starters LP not different between newly LP lower in export-stoppers (2003) Growth of LP higher for than for non-exporters prior entered and continuous exporters, than in continuous exporters, Nine sub-saharan exporters than for non- to entry. but higher compared to non- but higher than in non- African countries exporters. exporters. exporting firms. ( ) Mengistae and Pattillo TFP 17.4% higher on average, (2004) 18.6% for Kenya. Difference higher Three sub-saharan for direct exporters, insignificant for African countries indirect exporters. TFP growth on ( ) average 10% higher for exporters. Difference again higher for direct exporters, insign. for indirect exporters. Note: Studies are listed in alphabetical order for the countries covered and chronologically for each country (using the most recent version of the study). Studies covering up to three countries are listed separately for each country (if information on each country is available); multi-country studies covering more than three countries are listed at the end of the table. LP = labour productivity (total value of shipments per worker or value-added per worker. TFP = Total factor productivity, usually calculated as the residual from an estimated Cobb-Douglas-type production function. EP = exporter premia: cet. par. percentage difference of LP between exporters and non-exporters, usually based on OLS regressions controlling for industries, regions, firm size (no. of employees) and year. 27